This research article proposes accelerating a geodesic ray-tracing algorithm for fiber tracking in brain imaging using parallel programming on a GPU. Fiber tracking uses diffusion MRI to noninvasively examine brain fiber structures at a microscopic level. While geodesic ray-tracing is robust, it is computationally expensive to reliably find all fibers between seed points and target regions. The authors implemented a highly parallel version of the algorithm using NVIDIA's CUDA platform on a GPU. This provided a significant reduction in running time of up to 40x compared to a multithreaded CPU implementation, greatly increasing the applicability of the algorithm.